No abstract

No abstract

No abstract

Recently, radiography image elaboration using different image processing methods has been introduced as an alternative to enhance the radiographs. The ability of improving the quality of an image depends on the scattered x-ray and the acquisition data by electronic system in digital radiography (DR) (RT). Iterative methods, well known in general sparse signal reconstruction, can be suited for the radiography images. In this research, the DR image is improved by minimizing an objective function using the fast iterative shrinkage-thresholding algorithm (FISTA), Monotone FISTA (MFISTA), over relaxation MFISTA (OMFISTA) and converged FISTA, where the solution sparsity may be adjusted as desired. The paper surveys four well-known methods for sparse process, and assesses their optimization parameters with the goal of obtaining the best algorithm for industrial radiography images. First, the radiographs from the welded objects were provided and four iterative methods were implemented to the radiographs for enhancing the contrast. Then reconstructed images were assessed on the basis of their quality. The results show that the reconstructed images have better contrast than the original radiography and the OMFISTA method has a lower runtime compared to others. Also, the results demonstrate the viability and efficiency of the four proposed algorithms on radiography image deblurring problems without any information about the noise of radiography system.

Radiography is a technique used to view the internal defect regions in welded objects using x-rays or gamma rays. Information extraction from the radiographs depends on the image quality and the interpreter's skill. The use of image enhancement methods is an important goal in industrial radiographic testing for contrast improvement of radiographs. In this study, two cartoon-texture image decomposition methods (CTID) based on the nonlinear low pass and high pass filter (NLHF) algorithm and the total variation L1 (TV-L1) algorithm were used to detect and improve the defect(s) visualization in the radiographs of the GDXray database. These methods utilize iterative methods along with different filters to generate the cartoon and texture components of the image. The proposed methods were successfully implemented to the high and low contrast radiography images. Improvement in the defect regions was achieved while sharpening the edges and fine detail of the radiographs. Experts' evaluations showed that the contrast of the texture component in the NLHF method was better than the TV-L1 and the original radiograph.